ClawHub
Back to skill

Security audit

Uncle Matt's Build Me Something Anything

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed local project generator that may scan allowed project context, but it asks for boundaries and keeps writes contained to a fresh folder.

Install only if you are comfortable with an agent inspecting your allowed project files to infer patterns and create a new project. Set clear off-limits paths up front, avoid saying nothing is off-limits unless you mean it, and only opt into tool installs or ClawPatch if you trust that tooling.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
84% confidence
Finding
The skill claims a constrained purpose focused on local evidence scanning and generating a fresh-folder project, but these instructions expand behavior into optional use of external review tooling and possible installation workflows. That broadens the operational scope, introduces supply-chain and network exposure, and can cause the agent to take actions not clearly justified by the declared skill boundary.

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
Referencing npm metadata refresh and ClawPatch install/use implies interaction with external package sources and potentially installing or trusting third-party code during execution. In a skill whose stated purpose does not require package-source access, this creates unnecessary supply-chain risk and weakens the promised local-only/project-bounded behavior.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill explicitly instructs the agent to scan broadly across the user's local repo, sibling workspaces, mounted project roots, manifests, docs, filenames, tests, and assets before it has presented a clear upfront privacy warning in the primary behavior description. Although later sections add some boundary and approval language, the overall design still encourages expansive local discovery, which increases the risk of collecting or exposing sensitive project and personal information beyond what the user reasonably expects.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.